Market Research Agent / Evidence-backed intelligence

Chooses relevant tools, recalls prior evidence, and checks its conclusions. Follow the decisions behind every report.

Task-aware toolsPersistent RAGBounded feedback loop
  1. 01
    PlanUnderstand the question. Select the tools.
  2. 02
    ResearchCollect evidence. Recall prior news.
  3. 03
    AnalyzeCalculate signals. Build a thesis.
  4. 04
    VerifyCheck support. Retry if needed.

What would you like to investigate?

Examples fill the form. Research starts only when you click Run.

Prediction data: CoinRithm aggregation, original venue Polymarket. This is not Gamma direct data.

Model, horizon & advanced options

Off by default. Adds model calls when enabled.

Prediction data connection diagnostics
CoinRithm aggregation

Not checked. Run a connection check when needed; it does not start research.

Checking model configuration...
Latest quote
Fetching quote...Yahoo Finance

Price history

Loading daily prices…

Sector relative performance

Enter a stock to compare with its sector benchmark.

Replay a previous run

Loading saved reports...

The answer, with its evidence.

Waiting for input

Choose an example or ask your own question. Decisions, tool calls, and evidence will appear here.

Under the hood: the nine-node LangGraph workflow

The four stages group nine execution nodes. Evidence gaps return to targeted retrieval, with a shared retry budget. A completed report may still carry limitations.

Referenced evidence

Inspect the source behind this reference. A valid ID does not by itself prove that the source supports the claim.